A general ggplot template

ggplot(data = ..., aes(x = ..., y = ..., ...))+

geom_*()+ add geometrical objects (points, lines, areas, …)

facet_*()+ split in subplots

coord_*()+ changes the coordinate system

scale_*_*()+ changes the way geoms look

theme_*() changes the look & feel of the plot

base <- iris %>% 
  ggplot(aes(x = Sepal.Length, y = Sepal.Width, 
             color = Species, size = Petal.Width))+
  geom_point()
base

coord_* functions change the behavior / scale of the axes

• invert coordinates
• boundaries
• scale (linear, log)
• type (cartesian, polar)

base + coord_flip()

change the limits of the plot to ‘zoom’ on a part of it (!!handle with care!!)

base + coord_cartesian(xlim = c(5, 6), ylim = c(3,4))

sometimes you need the two scales on the two axes to be comparable

ggplot(mpg, aes(cty, hwy)) + geom_point() + labs(title = "It looks like mileage in City and Highway is the same!")

sometimes you need the two scales on the two axes to be comparable

ggplot(mpg, aes(cty, hwy)) + geom_point() + 
  coord_fixed() + labs(title = "But mileage is higher in highways!")

sometimes you need axes to be n an exact proportion

ggplot(mpg, aes(cty, hwy)) + geom_point() + 
  coord_fixed(ratio = 1/5) + labs(title = "City is 5 times as big as highway here")

log scales

ggplot(mpg, aes(cty, hwy)) + geom_point() + coord_trans(x = "log10", y = "log10")+ labs(title = "log axes")

we can also transform a (x,y) coordinate into a (distance, angle)

base + coord_polar()

this is the only way to make a pie chart in ggplot: a bar chart in polar coordinates.

ggplot(mpg, aes(x = "", y = class, fill = class)) + geom_col()

this is the only way to make a pie chart in ggplot: a bar chart in polar coordinates.

ggplot(mpg, aes(x = "", y = class, fill = class)) + geom_col() + coord_polar("y", start = 0)

scale_*_* functions change the appearance of the mapping

• general usage: scale_NAME_TYPE
• NAME can take values: alpha, color, fill, linetype, shape, size, x, y
• TYPE can take different values according to the scale type

make our base plot greyscale and change the size mapping

base + scale_color_grey() + scale_size_continuous(range = c(1, 10))

Manipulating colors is fun! but you need to pay attention to some stuff

• is your color variable continuous or discrete?
• do you want a gradient?
• can colorblind people use your plot?

Brewer palettes

base + scale_color_brewer()

Brewer palettes

base + scale_color_brewer(type = "qual")

Brewer has many palettes

base + scale_color_brewer(palette = "Set1")

Viridis palette

base + scale_color_viridis_d()

RColorBrewer::display.brewer.all()

base + scale_color_manual(values = c("pink", "violet", "brown"))

https://www.datanovia.com/en/blog/awesome-list-of-657-r-color-names/

a simple plot to showcase stuff

base2 <- ggplot(iris, aes(x = Sepal.Length, y = Sepal.Width, color = Petal.Width))+
  geom_point()
base2

change the gradient

base2 + scale_color_gradient(low = "green", high = "red")

a gradient with an intermediate step

base2 + scale_color_gradient2(low = "red", mid = "grey", high = "green", midpoint = mean(iris$Petal.Width))

Sometimes you have continuous data but you want to treat it as discrete: binned scales

base2 + scale_color_binned()

change the gradient to be colorblind proof

base2 + scale_color_viridis_c()

base + scale_size_continuous(range = c(1,12))

base + scale_size_binned(n.breaks = 3, range = c(1,12))

in barplots, the relevant variable is the fill

ggplot(mpg, aes(x = manufacturer, color = manufacturer)) + geom_bar()

in barplots, the relevant variable is the fill

prod <- ggplot(mpg, aes(x = manufacturer, fill = manufacturer)) + geom_bar()
prod

prod + scale_fill_viridis_d()

if appearance does not vary then pass it directly:

ggplot(mpg, aes(x = manufacturer)) + geom_bar(fill = "red")

theme functions refer to the appearance of everything but the mapping

• pre-installed themes (e.g. theme_minimal())
• extra themes (e.g. ggthemes, hrbrthemes)
• fine_tuning the themes: theme()
• changing the labesl: labs()

base + theme_classic()

base + theme_dark()

base + theme_linedraw()

base + theme_void()

install.packages("ggthemes")
library(ggthemes)
base + theme_excel_new()

library(hrbrthemes)
base + theme_ipsum_rc()

base + theme(legend.position = "left")

base + theme(panel.background = element_rect(fill = "yellow"))

base + labs(title = "title", subtitle = "subtitle", caption = "CAPTION", tag = "tag")

to save a plot to file, use ggsave(). By default it saves the last plot made

ggsave(filename, width, height, dpi)

sometimes you need to add some lements that are not in the data

• reference lines or areas
• text
• annotations

there are three types of lines:

• horizontal
• vertical
• y = ax + b

base + geom_hline(yintercept = 3, color = "red")

base + geom_vline(xintercept = 5, color = "blue")

base + geom_abline(slope = 0.8, intercept = -2, color = "orange")

there are two text geoms: text and label

carspeed <- mpg %>%  filter(year == 2008 & trans == "manual(m5)") %>% ggplot(aes(x = cty, y = hwy))
carspeed + geom_point()

labeling cars by their name: text

carspeed + geom_text(aes(label = model))

labeling cars by their name: label

carspeed + geom_label(aes(label = model))

just put a custom text somewhere

base + annotate(geom = "text", label = "ANNOTATION", x = 6, y = 4, hjust = 0)

• the ggplot cheatsheet is your friend (Help -> cheatsheets)
• stack overflow helps out for trickier questions
• not feeling inspired? 50 cool visualisations here: http://r-statistics.co/Top50-Ggplot2-Visualizations-MasterList-R-Code.html
• need a complete tour of possibilities? https://www.r-graph-gallery.com/